In this work, the influence of the speed of sound on the propagation of sound in the sea, and the possibilities for the improvement of the sound speed profile measurement accuracy were investigated. Based on this research, a new approach to the sound speed profile measurement was developed. A basic mathematical model for sound propagation in the sea, based on the ray theory, was introduced. With that model, the influence of the sound speed profile on the propagation of sound and their interdependence, were determined. The Adriatic Sea hydrographic data relevant for the sound speed determination (salinity, temperature) were studied. On the basis of the available results of numerous hydrographic and oceanographic measurements, extensive daily and seasonal, fluctuations of both, temperature and salinity are determined, which means that there are extensive fluctuations in the sound speed profile as well. The methods most frequently used for the sound speed measurements (sing-around, phase, STD) were analyzed, and their accuracy defined. For the sound speed profile measurement in the Adriatic Sea, the sing-around method was found to be optimal. Researching the influence of the speed of sound on the propagation of sound in the sea and analysing the existing measurement methods, a possibility for the measurement of the sound speed profile with an increased accuracy was discovered. The new measuring algorithm, named direct calibration and automated error correction algorithm was developed. It improves the measurement accuracy of the basic sing-around method. Applying proper calibration method, the parameters describing the causes responsible for the major measurement errors, are directly calculated. These parameters are thermal expansion coefficient (α), the acoustic signal total delay time ( τ ) and the acoustic base exact lenght (d). With the so calculated parameters, the measurement results are being automaticaly corrected in the real time, and the measurement accuracy is thus considerably increased. In the laboratory conditions, the total method accuracy depends only on the accuracy of the measurement of the water temperature in which sound speed measurement is done. With the water temperature measurement accuracy of 0.01°C, the total sound speed measurement accuracy better than 0.04 m/s can be attained in the whole measurement range. Special attention was paid to the possibility of applying this algorithm to the real time sound speed profile measurements in the Adriatic Sea. The possibility of applying the proposed algorithm in the dynamic conditions of the real environment was studied. In such conditions, the temperature changes in the material of the acoustic base would not follow the temperature changes recorded by the temperaure sensor on the profiler device. As a consequence, the automated error correction will not be proper. During the research, the time response of the temperature sensor and the acoustic base on the temperature change was determined, as well as their time constants. Also, the coefficient of temperature change was introduced as the descriptor of the temperature change the profiler device is subjected to during the measurement in the seawater. The coefficient includes the change due to temperature gradient of the seawater and the rate at which the device is being descended (duration of the measurement) during the measurement. A mathematical model of the dynamic measurement in the real seawater environment was also developed. The formulas directly linking the measurement accuracy with the coefficient of temperature change (temperature gradient of the seawater and the rate of the descending), were derived. With these formulas, the possibility of the measurement accuracy improvement in the dynamic conditions in the real seawater environment is demonstrated. The experimental confirmation on the built profiler device prototype, proved the accuracy improvement with the implementation of the proposed algorithm.